This paper presents an artificial neural network technique which is used for classification of multi-criteria inventory for retailers. The control of large inventory items is not possible for maximum profit with an equal attention. Generally, single criteria inventory classification is followed by inventory manager such as total cost. It has been realized that for retail organizations the single criteria inventory classification is less effective, So, instead of cost, there are some other criteria which are more important, which are profit per unit, demand of the item, shelf-life, of an item and lead time to the store. So a multi-criteria approach has been used here for inventory classification of retail outlets. For this purpose, the artificial neural network technique has been used the classification of inventory has carried out by pattern recognition and classification tool in MATLAB software. For training, the network scaled conjugate gradient backpropagation is used in this work. So, a classifier model is trained for the classification of multi-criteria inventory and prediction of inventory based on an expert system which can classify any number of items in retail outlets.